L

Learn AI

Track progress · learn offline

Open

Lesson 1 of 6

What even is an AI product?

6 min read

Your app has a 'summarize' button powered by AI. The one next door has an 'export to PDF' button. One of these is an AI product. What actually makes the difference?

Rules you write vs patterns it learns

Classic software runs on rules a human wrote: if the cart is over $50, add free shipping. Every path is spelled out in advance. An AI product flips that — the core behavior is learned from examples, not hand-coded. Nobody wrote the rule for 'summarize this messy email'; the model learned the pattern from mountains of text and applies it to inputs no one could have listed ahead of time.

An AI product's core behavior is learned, not written. That's what lets it handle open-ended, messy inputs a rules engine never could — it's narrow AI pointed at one job.

The trade you're making

Learned behavior buys you something huge — the ability to handle the unenumerable — but you pay for it. A rules engine is exact and predictable; a learned model is probabilistic. It will be wrong sometimes, it can't fully explain itself, and the same input can vary. Good AI product thinking starts by accepting that trade, not wishing it away: design for a system that's usually right, not always right.

You trade exact, predictable, and explainable for flexible, capable, and occasionally wrong. Whether that trade is worth it is a product decision — not a technical one.

Not every feature that touches a model is an 'AI product'. If a plain rule would do the job as well, use the rule — it's cheaper, faster, and never hallucinates. Reach for learned behavior when the input is too open-ended to enumerate.

The shape of it

A teammate says 'let's make it an AI product' for a form that emails a receipt when someone pays. What's the strategist's response?

Continue in the app

Take the whole AI Product Strategy course — tracked

Get your personalized path, progress and streaks in the app — this lesson and every next one, in order.